Residual Temperature Bias Effects in LIMS Stratospheric Ozone and Water Vapor

Abstract. The Nimbus 7 Limb Infrared Monitor of the Stratosphere (LIMS) instrument operated from October 25, 1978, through May 28, 1979. Its Version (V6) profiles were processed and archived in 2002. We present several diagnostic examples of the quality of the V6 stratospheric ozone and water vapor data based on their Level 3 zonal Fourier coefficient products. In particular, we show that there are small differences in the ascending (A) minus descending (D) orbital temperature-pressure or T(p) profiles (their A-D values) that affect (A-D) ozone and water vapor. Systematic A-D biases in T(p) can arise from small radiance biases and/or from viewing anomalies along orbits. There can also be (A-D) differences in T(p) due to not resolving and correcting for all of the atmospheric temperature gradient along LIMS tangent view-paths. An error in T(p) affects the retrievals of ozone and water vapor through: (1) the Planck blackbody function in forward calculations of limb radiance that are part of the iterative retrieval algorithm of LIMS, and (2) the registration of the measured LIMS species radiance profiles in pressure-altitude, particularly for the lower stratosphere. We evaluate V6 ozone profile biases in the upper stratosphere with the aid of comparisons against a monthly climatology of UV-ozone soundings from rocketsondes. We also provide results of time series analyses of V6 ozone, water vapor, and potential vorticity for the middle stratosphere to show that their average (A+D) V6 Level 3 products provide a clear picture of the evolution of those tracers during northern hemisphere winter. We recommend that researchers use the average V6 Level 3 data for their science studies of stratospheric ozone and water vapor wherever diurnal variations of them are unexpected. We also point out that the present-day Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) experiment is providing measurements and retrievals of temperature and ozone, which are essentially free of any anomalous diurnal variations.


1 Introduction and objectives 44 The historic Nimbus 7 Limb Infrared Monitor of the Stratosphere (LIMS) experiment provided Figure 3 shows V6 zonal average H2O for March 1979 from SPARC-DI. H2O is effectively a 158 tracer of the mean meridional circulation, which moves upward from the tropical tropopause to 159 the middle stratosphere and then poleward toward higher latitudes. Minimum values of H2O are 160 of order 3.5 ppmv in the tropics between 50 and 70 hPa. The sharply increasing H2O near the tropical tropopause is due, in part, to residual emissions from cirrus cloud tops that were not 162 screened completely from the bottom of the LIMS H2O radiance profiles prior to retrieval. 163 Highest values of H2O are at upper altitudes (> 6.0 ppmv) and are due to the oxidation of 164 methane (CH4) to H2O, followed by its net transport and accumulation at higher latitudes. NLTE 165 processes also cause enhancements of H2O radiance near the stratopause during daytime. Those 166 uncorrected NLTE effects extend downward to lower altitudes for retrieved V6 H2O, although 167 the effects are small for the middle and lower stratosphere (Mertens et al., 2002). Estimates of 168 the effect of temperature bias for V6 H2O are in Table 1 (row 4) from Remsberg et al. (2009). 169 170 3 Measurement, retrieval, and day/night differences for temperature 171 Nimbus 7 was in a near-polar orbit, and LIMS made measurements at ~1 pm local time along its 172 ascending (A or south-to-north traveling) orbital segments and at ~11 pm for its descending (D 173 or north-to-south traveling) segments. The A-D time difference is of the order of 10 hours 174 because LIMS viewed the atmosphere 146.5° clockwise of the spacecraft velocity vector or 33.5° 175 counterclockwise from its negative velocity vector, as seen from overhead. In other words, 176 LIMS viewed atmospheric tangent layers in opposing meridional directions for the NH and 177 through the tropics or toward the SSE along A segments and toward the NNW along D segments 178 (Gille and Russell, 1984). The A and D view paths for middle latitudes of the SH are more 179 nearly in a zonal direction and toward the NNW, respectively, due to the orbital inclination of how well the effects of the temperature tides have been resolved (Remsberg et al., 2004). 184 Tropical differences are due mainly to diurnal tides, and they become large in the mesosphere. 185 Tidal amplitudes for the tropics increase with altitude in Fig. 4, ranging from -2 K at 15 hPa to 186 +4 K at 1.5 hPa. Those V6 tidal variations agree qualitatively with ones from rocket Datasonde 187 profiles (Hitchman and Leovy, 1985;Finger et al., 1975). Fig. 4 also shows the expected 180° 188 change of phase for A-D T(p) from the tropics to subtropics. Accurate determinations of T(p) 189 versus latitude depend critically on knowledge of the Nimbus 7 spacecraft attitude. That information for a complete orbit comes empirically from profiles of calculated-to-measured 191 radiance ratios for the LIMS narrow CO2 channel and can lead to a bias error for A-D T(p). Any 192 bias in the orbital attitude will affect T(p) at all altitudes; that error source is small according to 193 the good comparisons of the LIMS-derived geopotential heights versus those from operational 194 analyses at both the 10-hPa and 46-hPa levels (Remsberg et al., 2004). Even so,Fig. 4 also 195 shows that there are residual A-D T(p) differences at 70 hPa that are opposite in sign at 40°S and 196 30°N, or just where there are large, opposing meridional gradients in T(p) in Fig. 2. 197 198 The measured ozone radiance profiles contain the full effects of any atmospheric variations in 199 T(p). As an example, Figure 5 shows zonal mean, ozone radiance differences (A-D) for one day  There are negative A-D ozone radiances of up to -5% in the stratosphere at the northern middle 207 latitudes, and they are a result of the meridional decrease of T(p) (in Fig. 2) from the northern 208 subtropics toward higher latitudes. More of the measured radiance in that region comes from the 209 front end of the tangent layer or from the colder side on the A orbital segment and from the 210 warmer side on the D segment, leading to negative A-D radiances. The LIMS algorithms for 211 temperature and species account for horizontal temperature gradients, to first order (Roewe et al., 212 1982;Gille et al., 1984;Remsberg et al., 2004 and2007). T(p) gradients for V6 are from daily 213 surface maps from the average (A+D) V5 temperature fields, where the meridional resolution of 214 the V5 fields is no better than half that of V6, or 4° versus 2° of latitude. Those average (A+D)  Po and T(p) undergo iteration until the calculated and measured, tangent layer radiances agree to 228 within the noise levels of the measured radiances over the pressure range of 2 to 20 hPa. Yet, the 229 noise value for the narrow CO2 channel is nearly 2% of the signal at 2 hPa. This level is where 230 the diurnal temperature tide has a larger amplitude and can impart a systematic, A-D bias in Po.

231
An A-D bias error in radiance is also significant; a calibration error of 1% causes a 0.6 K error in 232 T(p) for the middle and upper stratosphere (Remsberg et al., 2004, Table 3). Another possible 233 source of (A-D) bias for T(p) can arise from a residual uncertainty of the viewing attitude of 234 LIMS along an orbit, its empirical "twist factor". 235 236 Roewe et al. (1982) showed that adjustments for horizontal gradients in T(p) affect species 237 retrievals from calculations of the Planck blackbody radiance throughout the stratosphere, as shown) increases from 40°N to 18°N, holds nearly steady in the tropics, and decreases from 20°S 244 to 40°S, mainly due to the changing ozone with latitude ( Fig. 1). Gordley and Russell (1981) 245 showed that the bulk of the LIMS broadband ozone radiance for the middle and lower 246 stratosphere also comes from the near side of the tangent layer (displaced toward the satellite by 247 about 300 to 500 km or ~3° to 6° of latitude). Such tangent layer asymmetries explain part of the 248 observed change of sign of the A-D radiances between the two hemispheres in Fig. 5.
Nevertheless, the mass path algorithm of the V6 forward model simulates radiance along a well-250 resolved limb path, using rigorous ray tracing methods, including refraction effects and first-251 order corrections for temperature gradients, and assigns an observed tangent altitude 252 corresponding to the center of the measurement field-of-view. 253 254 Roewe et al. (1982) showed that adjustments for the path gradients of the ozone mixing ratio 255 itself imparts only small A-D mixing ratio differences (~2%). Thus, the V6 retrievals do not 256 account for species gradients. The V6 algorithms are no longer operational for further studies of 257 the effects of T(p) gradients on ozone and H2O. Instead, in the next section we present 258 diagnostic plots based on the V6 Level 3 data themselves to indicate that there are residual biases 259 in the distributions of V6 T(p) and that they carry over to V6 ozone and H2O. The V6 H2O retrievals are more sensitive than ozone to biases in T(p) at 3 hPa (in Table 1) 298 because most of the V6 H2O radiance comes from its strong, nearly saturated lines. Figure 9 299 shows the H2O A-D mixing ratio values for March. Both species are altered by horizontal 300 gradients in T(p) in the same way in calculations of their Planck radiances. The locus of 301 maximum percentage difference for H2O in the SH differs from that of ozone (Fig. 6) in the 302 middle to upper stratosphere because their respective mixing ratios also have gradients that 303 differ. The effect of the tropical temperature tide on H2O is not apparent at 1.5 hPa because of 304 the excess of NLTE radiances of H2O at and above that level during daytime.

Middle and lower stratosphere
307 V6 A-D ozone mixing ratio in Fig. 6 is near zero at 20 hPa. This feature occurs where V6 A-D 308 for T(p) in Fig. 4 is also small, or where there is iteration of Po and from which a hydrostatic 309 integration occurs both above and below that level. The ozone differences become negative 310 below that level across the tropics and in the NH, where the vertical gradient of ozone (Fig. 2) is 311 large and subject to small A-D differences in the registration of the ozone radiance profiles.

312
However, the ozone differences at SH middle latitudes remain positive down to the 100-hPa 313 level; only tangent views along the descending orbital path are in a nearly meridional direction at 314 those latitudes. In particular, the A-D ozone values in Fig. 6 are rather large at 40°S and 30°N 315 (40 to 100 hPa), and they are opposite in sign to the A-D T(p) differences of order ±1 K in Fig. 4.

316
This finding agrees with the estimates of T(p) effects at 50 hPa in Table 1, where a bias of -1.3 K 317 leads to a +20% bias in ozone. The A-D temperature biases are large just where the meridional 318 temperature gradients are also large (Fig. 2) and corrections for them are too small.   Table 1). The ROCOZ profiles are averages of the three April soundings for 1976-393 1978, and the horizontal bars at 0.5, 1.5, 3, 7, and 15 hPa are their estimated uncertainty of <10% (or <7% for ozone number density versus altitude, plus <3% for the conversion to mixing ratio 395 versus pressure, as taken from Table II -7 of Krueger (1984)). Fig. 13 indicates agreement to 396 within the estimates of bias error for V6 ozone at most altitudes of the stratosphere. V6 ozone is 397 higher than ROCOZ ozone from ~2.0 to 0.3 hPa. mesosphere. The increasingly positive, V6 day minus ROCOZ differences from winter to late 403 spring in the lower mesosphere are due to uncorrected NLTE emissions from CO2 and ozone that 404 increase toward lower solar zenith angles (Edwards et al., 1996;Manuilova et al., 1998). On the 405 other hand, the V6 daytime ozone of April and May is also larger than ROCOZ ozone in the 406 uppermost stratosphere, where NLTE should not be an issue. This finding implies that there may 407 be a slight negative bias for V6 T(p) at those high altitudes (see Table 1). Another possibility is 408 that the limited ROCOZ climatology at Wallops Islands may not be truly representative of zonal  Ozone is an effective tracer of the transport of air in and around the winter polar vortex on the 435 850 K surface (~10 hPa) (Leovy et al., 1985). Ozone also varies nearly monotonically at this  Water vapor is also a tracer of the net transport in the middle stratosphere. Figure 17 shows the